Exploring applications of generative AI and multi-modal learning in human-robot interaction, with a focus on intent prediction and decision-making systems.
Developing efficient and accurate 3D gaze tracking systems for automotive applications and human-computer interaction, with a focus on appearance-based methods.
Researching advanced techniques for object registration and pose estimation in mixed reality and robotic applications, with expertise in real-time systems.
Conducting research at CAIRO HCI Lab under Prof. Bradley Hayes on enhancing human intent recognition using Partially Observable Markov Decision Processes (POMDP)-based algorithms and integrating vision-based learning techniques with the Sawyer Robotic Manipulator.
Optimized the I2D-Net Eye Gaze Estimation model, reducing parameters by 0.5x and achieving 2x inference speed, enabling faster gaze tracking for interactive automotive HUDs. Collaborated with Faurecia to extend interactive automotive heads-up display features with gaze-based vs gesture-based on-road distraction detection.
Partnered with Collins Aerospace to develop a mixed reality-based assembly system using custom-trained object detection models with tailored eye gaze and hand tracking for efficient, real-time assembly guidance. Developed a robust motion tracking framework with a markerless ByteTrack model, achieving 97.8% mAP detection accuracy.